Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
https://doi.org/10.1590/1980-549720210045
ORIGINAL ARTICLE / ARTIGO ORIGINAL
Impact of the Dry Law on road traffic
mortality in Brazilian states: an interrupted
time series analysis
Impacto da Lei Seca sobre a mortalidade no trânsito nas unidades
federativas do Brasil: uma análise de série temporal interrompida
Hélio Rubens de Carvalho NunesI , Cristiane Murta-NascimentoI , Maria Cristina
Pereira LimaII
ABSTRACT: Objective: To assess the impact of 2008 Public Law number 11,705, also known as Dry Law
(DL-08), on mortality from road traffic accidents (RTA), in each of the 27 Brazilian Federative Units (BFUs).
Methods: Ecological study of interrupted time series with RTA data from 2002 to 2015, totalizing 168 months.
Data were obtained from the Mortality Information System, the Brazilian Institute of Geography and Statistics,
and from the National Traffic Department. Autoregressive integrated moving average (ARIMA) models were
adjusted to assess the impact of DL-08 in each BFUs. Results: After the implementation of the DL-08, there
was a significant decrease in mortality from RTA in the state of Santa Catarina (pre DL-08 = 2.60 ± 0.30 and
post DL-08 = 2.32 ± 0.35; p < 0.001) and in the Federal District (pre DL-08 = 2.22 ± 0.40 and post DL-08 = 1.76
± 0.35; p = 0.002), a significant increase in mortality in the states of Acre, Amazonas, Rondônia, Maranhão,
Piauí, Ceará, Rio Grande do Norte, Paraíba, Pernambuco, Alagoas, Sergipe and Mato Grosso, and a stability
in the other states. The sensitivity analysis conducted over a shorter time series with 24 months showed results
similar to those obtained with the 168-month series for most of the 27 BFUs. Conclusion: The DL-08 had a
heterogeneous impact on mortality from traffic accidents on BFUs.
Keywords: Interrupted time series analysis. Accidents, traffic. Time series studies. Ecological studies.
I
Department of Public Health, Medical School of Botucatu, Universidade Estadual Paulista “Júlio de Mesquita Filho” – Botucatu
(SP), Brazil.
II
Department of Neurology, Psychology and Psychiatry, Medical School of Botucatu, Universidade Estadual Paulista “Júlio de
Mesquita Filho” – Botucatu (SP), Brazil.
Corresponding author: Hélio Rubens de Carvalho Nunes. Faculdade de Medicina de Botucatu, Universidade Estadual Paulista
“Júlio de Mesquita Filho”. Av. Prof. Mário Rubens Guimarães Montenegro, s/n., Vila Paraíso, CEP: 18618-687, Botucatu, SP, Brazil.
E-mail: hrcn@outlook.com.br
Conflict of interests: nothing to declare – Financial support: none
1
REV BRAS EPIDEMIOL 2021; 24: E210045NUNES, H.R.C. ET AL.
RESUMO: Objetivo: Analisar o impacto da Lei 11.705, conhecida por “Lei Seca” (LS-08), sobre a mortalidade por
acidentes de trânsito (AT) em cada uma das 27 unidades federativas (UF) do Brasil. Método: Estudo ecológico
de séries temporais interrompidas com dados de AT entre 2002 a 2015, totalizando 168 meses. Os dados foram
obtidos do Sistema de Informações sobre Mortalidade, do Instituto Brasileiro de Geografia e Estatística e do
Departamento Nacional de Trânsito. Foram ajustados modelos auto-regressivos integrados de médias móveis
(ARIMA) para analisar o impacto da LS-08 em cada UF. Resultados: Após a implantação da LS-08, a mortalidade
por AT diminuiu significativamente no estado de Santa Catarina (pré-LS-08 = 2,60 ± 0,30 e pós-LS-08 = 2,32
± 0,35; p < 0,001) e no Distrito Federal (pré-LS-08 = 2,22 ± 0,40 e pós-LS-08 = 1,76 ± 0,35; p = 0,002), aumentou
significativamente nos estados do Acre, Amazonas, Rondônia, Maranhão, Piauí, Ceará, Rio Grande do Norte,
Paraíba, Pernambuco, Alagoas, Sergipe e Mato Grosso e permaneceu estável nos demais. Análise de sensibilidade
conduzida sob uma série temporal mais curta, com 24 meses, apresentou resultados semelhantes aos obtidos
com a série de 168 meses para a maioria das 27 UF. Conclusão: A LS-08 exerceu impacto heterogêneo sobre a
mortalidade por AT entre as UF.
Palavras-chave: Análise de séries temporais interrompida. Acidentes de trânsito. Estudos de séries temporais.
Estudos ecológicos.
INTRODUCTION
Traffic accidents (TA) are a relevant public health problem. They account for about 1.35 mil-
lion deaths per year worldwide, being the eighth leading cause of death in the general popu-
lation and the leading cause of death among people aged 5 to 29 years1. Furthermore, most
TAs occur in low- and middle-income countries, where their cost can reach 2.9% of the
Gross Domestic Product (GDP)2.
In Brazil, data from the 2013 National Health Survey show that 3.1% of the Brazilian
population were involved in TAs with bodily injuries within a 12-month interval3, which
places TA as the second cause with the highest proportion of years of life lost due to exter-
nal causes in Brazil4.
The effects of alcohol consumption on the occurrence of TAs are well described in
the literature. Alcohol is known to delay the reaction to a stimulus, alter alertness, visual
acuity, performance in tasks that involve divided attention and the ability to judge5-7.
Furthermore, studies show that alcohol intake is associated with a higher risk of injuries
and death from TA8,9. In Brazil, Damacena et al.10 showed that alcohol consumption is asso-
ciated with TA even when the effects of age, sex, education, skin color and marital status
are considered. Evidence of the harmful effects of alcohol presented in the world litera-
ture helps to understand why 45 countries currently have restrictive laws on alcohol and
driving1, including Brazil.
In Brazil, Law No. 11,705, of June 19, 2008, popularly known as Dry Law (DL-08),
was one of the measures adopted by the Brazilian government to reduce morbidity and
2
REV BRAS EPIDEMIOL 2021; 24: E210045Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
mortality from TA. In force since June 20, 2008, the DL-08 penalizes the driver caught driv-
ing under the influence of alcohol or any psychoactive substance with the suspension of
the right to drive for 12 months across Brazilian territory, in addition to fines and admin-
istrative sanctions11.
Several studies have investigated the impact of DL-08 on TA mortality in Brazil.
Malta et al.12 did this by comparing the overlapping confidence intervals of the standard-
ized rate of mortality from TA before and after DL-08, using aggregated data from one
year before and one year after the law implementation. The authors reported a signifi-
cant reduction (-7.4%) in the standardized rate of mortality from TA in Brazil and also
stated that this reduction occurred heterogeneously across states. Andreuccetti et al.13
showed that DL-08 was responsible for a significant reduction in the rate of mortality
from TA by 7.2% and 16.0% in the state of São Paulo and capital, respectively. Abreu
et al.14, in a descriptive study in the city of Rio de Janeiro, compared the months of July
2007 (pre-DL-08) and July 2008 (post-DL-08) and reported a reduction of 12.9% in mor-
tality from TA. In 2017, Klabunde et al.15 compared pre- and post-DL-08 TA mortality
for the state of Santa Catarina and reported a significant reduction in the overall rate.
More recently, Abreu et al.16, in a segmented regression with yearly data for the state of
Paraná between 1980 and 2014, observed a decrease in mortality among drivers in gen-
eral and pedestrians after DL-08.
Despite the relevance of studies already carried out, the effect of DL-08 has not yet been
evaluated by models that simultaneously incorporate autocorrelation, seasonality, con-
founders and cointerventions in all states. Thus, TA mortality described by closer-to-real-
ity and regionalized models can provide more accurate estimates of the impact of DL-08,
contributing to traffic management in the country. Therefore, the aim of this study was to
assess the impact of DL-08 on mortality from TA in the 27 Brazilian Federation Units (UF),
between January 2002 and December 2015.
METHODS
This study conducted an interrupted time-series analysis (ITSA) to assess mortality from
TA that occurred monthly in each of the 27 Brazilian FUs between 2002 and 2015, in which
168 months were included, being 78 months before DL-08 ( January 2002 to June 2008) and
90 after DL-08 ( July 2008 to December 2015).
The studied outcome was the mortality coefficient due to TA observed monthly in
each FU. Deaths whose underlying causes were coded in categories V01 to V89 were used
for calculation, representing land transport accidents in Chapter XX (External Causes
of Morbidity and Mortality) of the tenth revision of the International Classification of
Diseases (ICD-10).
Three sources of information were used to collect data: Mortality Information System17
for the monthly number of deaths from TA; Brazilian Institute of Geography and Statistics
3
REV BRAS EPIDEMIOL 2021; 24: E210045NUNES, H.R.C. ET AL.
(IBGE)18 to collect the annual population projection; and the National Traffic Department
(DENATRAN)19 for information about the monthly vehicle fleet.
The intervention studied was DL-08, which was established in Brazil on June 20, 2008.
Two co-interventions were also considered: the implementation of the Emergency and
Emergency Medical Service (SAMU), according to Ordinance 5,055 of April 27, 2004 (SAMU-
2004), and its reformulation according to Ordinance of the Ministry of Health 1,010 of May
21, 2012 (SAMU-2012).
Regressors considered were:
• seasonality, represented by 11 polychotomous variables, one for each month (from
February to December) and assuming values of -1 for January (reference month), 1
for the current month and 0 otherwise;
• number of Saturdays and Sundays per month;
• state fleet of motor vehicles.
A model of the type Zt = μ + w0It + Xtβ + Nt was adjusted for each FU, with μ being a
constant, Zt the TA mortality coefficient in month t (t=1...168), w0 the impact of DL-08, It a
binary variable equal to 0 when t ≤ 78 and equal to 1 when t > 78, Xt is the regressor matrix
and Nt a residual series belonging to the autoregressive integrated moving average ARIMA
(p,d,q). To identify the residual Nt series and to verify the assumptions of homoscedasticity
and stationarity, the 78 pre-DL-08 observations were grouped into 13 clusters containing
six observations in temporal sequence.
Homoscedasticity was verified with a scatter plot between means and standard devia-
tions of the clusters. In the presence of heteroscedasticity, a Box-Cox transformation was
applied to the data.
Stationarity was verified by comparing the series:
• of original (or transformed) data;
• of the first difference;
• of the second difference, in relation to the behavior displayed in the graphs of time
series and the autocorrelation function (ACF), in which a decay of autocorrelations
was observed.
In case of doubt about which of the three series was stationary, the increased Duckey-
Fuller test was applied, and its regression equation was determined based on the number
of significantly non-null partial ACF.
The orders p and q were determined by comparing models with different orders p and
q in relation to the value of the Akaike information criterion (AIC). The need for a con-
stant in the model was verified with the t test for H0: μ = 0. The adequacy of the model
was evaluated by graphic inspection of residuals using the ACF function, partial ACF func-
tion and residual histogram, and by the Ljung-Box test. Analyses were performed using the
TSA, lmtest and AID packages of the R software, version 3.4.2.
4
REV BRAS EPIDEMIOL 2021; 24: E210045Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
RESULTS
Table 1 shows the estimates of the impact of DL-08 on traffic mortality, in each FU,
based on the time series with 168 months, being 78 months before DL-08 and 90 months
after DL-08, adjusted by cointerventions and regressors. A significant decrease in mor-
tality from TA after DL-08 was observed only in the state of Santa Catarina (w^0 = -0.08;
p < 0.001) and in the Federal District (w^0 = -0.08; p = 0.002). There was also a signifi-
cant increase in mortality from TA after DL-08 in 12 Brazilian states: Acre (w^0 = 0.22;
p = 0.005), Amazonas (w^0 = 0.08; p = 0.030) and Rondônia (w^0 = 0.71 ; p < 0.001) in the
North region; Alagoas (w^0 = 0.31; p = 0.008), Ceará (w^0 = 0.35; p < 0.001), Maranhão
(w^0 = 0.22; p = 0.004), Paraíba (w^0 = 0.13; p < 0.001), Pernambuco (w^0 = 0.30; p < 0.001),
Piauí (w^0 = 0.34; p = 0.007), Rio Grande do Norte (w^0 = 0.25; p < 0.001) and Sergipe
(w^0 = 0.30; p < 0.001) in the Northeast region; and Mato Grosso (w^0 = 0.41; p < 0.001)
in the Mid-west region. In the other FUs, mortality from TA after DL-08 remained stable
when compared to the pre-DL-08 period. The quality of adjusted models was evaluated
using residuals and the Ljung-Box test, which did not indicate a failure in the white noise
assumption for the residuals in all FUs.
Table 2 shows the sensitivity analysis performed to compare the results of the original
168-month time series with the results of the shorter 48-month time series, 24 months
pre-DL-08 and 24 months post-DL-08. Among the 27 UF, the impact of DL-08 (w^0) was
significant, having presented the same direction in both time series of six UF (Santa
Catarina, Federal District, Rondônia, Pernambuco, Sergipe and Mato Grosso), while
w^0 was not significant in both the time series of 10 UF (Amapá, Pará, Roraima, Bahia,
Mato Grosso do Sul, Espírito Santo, Minas Gerais, Rio de Janeiro, São Paulo and Rio
Grande do Sul), w^0 had the same direction with significant p value only in the longer
time series in five FUs (Acre, Ceará, Maranhão, Paraíba and Rio Grande do Norte), w^0
presented opposite directions and significant p value only in the longer time series in
three FUs (Amazonas, Alagoas and Piauí), w^0 had the same direction and significant p
value only in the shorter time series in two FUs (Goiás and Paraná) and, finally, w^0 pre-
sented opposite directions between series and significant p value only in the shorter
series in one FU (Tocantins).
DISCUSSION
This study found that the DL-08 was not enough to reduce TA mortality in all FUs.
After DL-08 implementation, mortality from TA decreased significantly in two states
(Santa Catarina and Distrito Federal), decreased non-significantly in five states (Amapá,
Rio de Janeiro, São Paulo, Paraná and Rio Grande do Sul), increased non-significantly in
eight states (Roraima, Pará, Tocantins, Bahia, Goiás, Mato Grosso do Sul, Minas Gerais
and Espírito Santo) and increased with statistical significance in 12 states (Amazonas, Acre,
5
REV BRAS EPIDEMIOL 2021; 24: E210045NUNES, H.R.C. ET AL.
Table 1. Estimates of the impact of the Dry Law (DL-08) on the monthly coefficient of mortality
from traffic accidents in each federative unit in Brazil.
Pre DL-08 Post DL-08 DL-08
FU Nt* p value&
(n = 78) (n = 90) w0# (p$)
North
Acre 1.22 ± 0.45 1.50 ± 0.57 ARMA (3,2) 0.22 (0.005) 0.345
Amapá 1.56 ± 0.62 1.42 ± 0.62 ARMA (2,2) -0.16 (0.068) 0.191
Amazonas 0.91 ± 0.24 1.00 ± 0.23 MA (1) 0.08 (0.030) 0.231
Pará 1.10 ± 0.20 1.51 ± 0.31 AR (1) 0.07 (0.204) 0.195
Rondônia 1.96 ± 0.47 2.75 ± 0.67 ARMA (1,1) 0.71 (< 0.001) 0.691
Roraima 2.25 ± 0.99 2.53 ± 0.99 IMA (1,1) 0.08 (0.798) 0.677
Tocantins 2.51 ± 0.62 3.00 ± 0.68 ARMA (3,3) 0.18 (0.315) 0.645
Northeast
Alagoas 1.61 ± 0.41 1.97 ± 0.43 ARMA (1,1) 0.31 (0.008) 0.771
Bahia 0.93 ± 0.19 1.37 ± 0.25 IMA (1,1) 0.01 (0.948) 0.216
Ceará 1.67 ± 0.23 2.08 ± 0.40 ARMA (1,1) 0.35 (< 0.001) 0.146
Maranhão 1.09 ± 0.25 1.85 ± 0.39 MA (1) 0.22 (0.004) 0.817
Paraíba 1.49 ± 0.32 1.95 ± 0.39 AR (1) 0.13 (< 0.001) 0.174
Pernambuco 1.41 ± 0.18 1.73 ± 0.23 ARMA (2,3) 0.30 (< 0.001) 0.252
Piauí 1.80 ± 0.44 2.93 ± 0.59 AR (1) 0.34 (0.007) 0.467
Rio Grande do Norte 1.20 ± 0.27 1.46 ± 0.28 MA (0,0,1) 0.25 (< 0.001) 0.994
Sergipe 1.77 ± 0.48 2.26 ± 0.47 ARMA (2,2) 0.30 (< 0.001) 0.163
Mid-West
Distrito Federal 2.22 ± 0.40 1.76 ± 0.35 MA (1) -0.08 (0.002) 0.326
Goiás 2.21 ± 0.33 2.52 ± 0.35 ARMA (1,1) 0.20 (0.085) 0.074
Mato Grosso 2.59 ± 0.46 3.02 ± 0.48 MA (2) 0.41 (< 0.001) 0.760
Mato Grosso do Sul 2.44 ± 0.42 2.59 ± 0.47 ARMA (1,1) 0.04 (0.741) 0.910
Southeast
Espírito Santo 2.28 ± 0.38 2.34 ± 0.38 AR (2) 0.003 (0.972) 0.410
Minas Gerais 1.48 ± 0.22 1.75 ± 0.19 IMA (1,1) 0.01 (0.904) 0.503
Rio de Janeiro 1.55 ± 0.18 1.38 ± 0.19 ARIMA (2,1,1) -0.10 (0.341) 0.256
São Paulo 1.46 ± 0.13 1.34 ± 0.15 ARMA (4,7) -0.06 (0.627) 0.111
Continue...
6
REV BRAS EPIDEMIOL 2021; 24: E210045Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
Table 1. Continuation.
Pre DL-08 Post DL-08 DL-08
FU Nt* p value&
(n = 78) (n = 90) w0# (p$)
South
Paraná 2.43 ± 0.30 2.46 ± 0.33 IMA (0,1,2) -0.01 (0.894) 0.119
Rio Grande do Sul 1.58 ± 0.18 1.52 ± 0.19 IMA (1,1) -0.18 (0.060) 0.295
Santa Catarina 2.60 ± 0.30 2.32 ± 0.35 ARMA (2,2) -0.08 (< 0.001) 0.921
*Residual series adjusted by models of the autoregressive integrated moving average (ARIMA); #estimate of the impact
of DL-08 seasonally adjusted, co-interventions SAMU-2004, SAMU-2012, state vehicle fleet and number of Saturdays
and Sundays in the month and autocorrelation; $p regarding the impact of DL-08 on mortality from TA; &p referring to
the Ljung-Box test on residuals of adjusted models.
Rondônia, Maranhão, Piauí, Ceará, Rio Grande do Norte, Paraíba, Pernambuco, Alagoas,
Sergipe and Mato Grosso).
Although interventions that regulate and penalize drinking and driving are among the
most effective to reduce mortality from TA20, this study pointed out a significant decrease
in mortality from TA after DL-08 in only two of the 27 FUs.
To explain this divergence, factors associated with TA deaths should be considered, par-
ticularly vehicle fleet, behavior of drivers, functioning of the national traffic system and
traffic engineering interventions, traffic municipalization process and adherence to health
policies and actions, which may have evolved differently between states.
As for the vehicle fleet, the study showed a direct linear correlation between the
fleet growth rate and the magnitude of the DL-08 effect (w^ 0) between FUs (r = 0.55;
p = 0.003; Spearman’s correlation). Thus, FUs with high positive variations in the
number of automotive vehicles between 2002 and 2015 had higher values of w^ 0,
which indicates that the effect of DL-08 may have lost strength in FUs whose popu-
lation acquired more vehicles. The high interest in purchasing vehicles is described
in the literature. Carvalho and Pereira 21 showed that families spend, on average, five
times more on private transportation than on public transportation. Public transport
has been increasingly ignored and this has an impact on urban mobility. From 1992
to 2012, commuting time between residence and workplace increased on average by
6.4% in Brazil 22.
With regard to behaviors associated with mortality from TA, such as those described by
Nascimento and Menandro23—alcohol and drug consumption, high-speed driving, disobedi-
ence to traffic rules and overestimation of driving skills—, the prevalence may have evolved
differently depending on state during the period of this study. The same may have occurred
with non-behavioral factors associated with mortality from TA, such as those mentioned
by Almeida et al.24 and Lima et al.25—age of the fleet, degree of urban development, traf-
fic engineering conditions and meteorological conditions—, which may also have evolved
differently between UFs.
7
REV BRAS EPIDEMIOL 2021; 24: E210045NUNES, H.R.C. ET AL.
Table 2. Results of the sensitivity analysis assessing the impact of DL-08 based on the longest
original time series (168 months) and on a shorter time series (48 months).
Impact of DL-08
48-month series 168-month series
w0* (p value#) w0$ (p value&)
^ in the same direction and significant p value in both time series
W0
Santa Catarina -0.09 (< 0.001) -0.08 (< 0.001)
Distrito Federal -0.07 (0.005) -0.08 (0.002)
Rondônia 0.60 (0.001) 0.71 (< 0.001)
Pernambuco 0.19 (< 0.001) 0.30 (< 0.001)
Sergipe 0.28 (< 0.001) 0.30 (< 0.001)
Mato Grosso 0.51 (< 0.001) 0.41 (< 0.001)
Stable TA mortality after DL-08 in both time series
Amapá 0.03 (0.806) -0.16 (0.068)
Pará 0.19 (0.202) 0.07 (0.204)
Roraima -0.13 (0.400) 0.08 (0.798)
Bahia 0.002 (0.983) 0.01 (0.948)
Mato Grosso do Sul 0.01 (0.892) 0.04 (0.741)
Espírito Santo -0.21 (0.222) 0.003 (0.972)
Minas Gerais 0.02 (0.757) 0.01 (0.904)
Rio de Janeiro -0.14 (0.165) -0.10 (0.341)
São Paulo -0.06 (0.634) -0.06 (0.627)
Rio Grande do Sul -0.22 (0.053) -0.18 (0.060)
^ in the same direction and significant p value only in the longer time series
W0
Acre 0.15 (0.123) 0.22 (0.005)
Ceará 0.08 (0.480) 0.35 (< 0.001)
Maranhão 0.22 (0.051) 0.22 (0.004)
Paraíba 0.06 (0.138) 0.13 (< 0.001)
Rio Grande do Norte 0.10 (0.189) 0.25 (< 0.001)
^ in opposite directions and significant p value only in the longer time series
W0
Amazonas -0.09 (0.153) 0.08 (0.030)
Alagoas -0.03 (0.780) 0.31 (0.008)
Piauí -0.09 (0.631) 0.34 (0.007)
Continue...
8
REV BRAS EPIDEMIOL 2021; 24: E210045Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
Table 2. Continuation.
Impact of DL-08
48-month series 168-month series
w0* (p value ) #
w0$ (p value&)
^ in the same direction and significant p value only in the shorter time series
W0
Goiás 0.29 (NUNES, H.R.C. ET AL.
Finally, differences in road structures, in the operationalization of traffic engineering
and in the assertiveness of works over the years between FUs can also contribute to the
variability observed.
The results of the sensitivity analysis obtained over the shorter time series of
24 months suggest that, in the states of Santa Catarina, Distrito Federal, Rondônia,
Pernambuco, Sergipe and Mato Grosso, there was a significant decrease or increase in
mortality from TA after DL-08 in the short term and extended to the long term. In the
states of Acre, Ceará, Maranhão, Paraíba and Rio Grande do Norte, a possible effect
of the time series size on the power of the tests in adjusted models must be consid-
ered. In the states of Amazonas, Alagoas and Piauí, the effect of DL-08 may have been
important only in the short term, with the historical trend of increased mortality from
TA prevailing, while in the state of Paraná, DL-08 may have lost efficiency over time.
However, the possibility that DL-08 has maintained its long-term efficiency in the state
of Paraná cannot be ruled out, but other factors associated with mortality from TA
have increased in the long-term.
The possible biases of this study were discussed using the instrument Risk Of Bias In
Non-randomised Studies of Interventions (ROBINS-I)31.
We believe that the risk of confounding bias is low because DL-08 is a federal law, which
submits any driver to its rules regardless of their predictors of mortality from TA.
In this study, for selection bias, one would need:
• an association between intra- and inter-regional migrations and DL-08, in
such a way that the demographic composition would be different in pre- and
post-DL-08 phases;
• a change in demographic composition accompanied by a change in behavioral and
non-behavioral factors associated with mortality from TA. As the probability of
these two conditions having occurred is subjectively small, we believe that the risk
of bias is low.
As for the risk of bias in the classification of interventions, we believe that this risk is
null because there was no possibility that the inhabitants of all 27 states were classified as
not exposed to DL-08 when it was in force, nor they were considered exposed to DL-08
before its validity.
In this study, with aggregated data from all FUs, the problem of missing data is discussed
in the context of international emigration. Data from the 2010 Demographic Census indi-
cate that the percentage of people who moved to other countries ranged from 0.06% in
Piauí to 0.57% in Goiás. Therefore, the percentage of people who leave Brazil is insufficient
to impact the estimates of monthly mortality coefficients for TA. Thus, we believe the risk
of bias due to lost data is low or null.
In addition, this study has advantages over previous work on the impact of DL-08 12-
16
. This is the first study to investigate the effect of DL-08 in all FUs using a statistical
10
REV BRAS EPIDEMIOL 2021; 24: E210045Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
intervention model that simultaneously considers ACF in monthly observations, sea-
sonality and the following confounding variables: number of Saturdays and Sundays
per month, evolution of vehicle fleet in each UF and co-interventions that represent
the implementation of SAMU, according to Ordinance 5,055 of April 27, 2004 (SAMU-
2004), and its reformulation in the Ministry of Health Ordinance 1,010 of 21 of May
2012 (SAMU-2012).
Time limits for pre- and post-DL-08 periods were determined based on Penfold and
Zhang32, who recommend at least eight observations before and eight observations after
an intervention. However, considering that our model would include several regressors
and that the TA mortality time series showed great variability in some FUs, we decided to
expand our time limit from January 2002 to December 2015 (n = 168 months), expanding
the recommended by Penfold and Zhang32.
However, this study has some limitations:
• other co-interventions could have been considered, including the National Traffic
Policy of September 2004 and its reformulation, approved in December 2014;
• alternative transfer functions for the impact of DL-08 could have been tested;
• evolution covariates of behavioral and non-behavioral factors associated with TA
could not be considered due to lack of secondary data;
• alcohol consumption may also have evolved differently between FUs.
Unfortunately, data on the annual prevalence of alcohol consumption were not
found for all 27 FUs. One possibility would be to use data from the Surveillance of Risk
Factors and Protection for Chronic Diseases by Telephone Survey (Vigitel), which pro-
vides an estimate of proportion of the number of adults who abused alcohol. However,
the first edition of Vigitel is from 2006 (therefore, it does not cover the period of our
study—2002 to 2015). Also, these data represent the capitals of each UF only, not the
UF as a whole.
In summary, after DL-08 there was a significant decrease in mortality from TA only in
the state of Santa Catarina and in the Federal District. This heterogeneity may stem from
factors associated with death from TA such as change in vehicle fleet, change in behavior of
vehicle drivers, functioning of the national traffic system, municipalization of traffic man-
agement and variable adherence to health policies and actions, which evolved unevenly
between the FUs.
ACKNOWLEDGEMENTS
We would like to thank the employees of the Collective Health Research Unit (UPeSC)
of the Botucatu Medical School, UNESP.
11
REV BRAS EPIDEMIOL 2021; 24: E210045NUNES, H.R.C. ET AL.
REFERENCES
1. World Health Organization. Global status report on abusivo de álcool e envolvimento em acidentes de
road safety 2018 [Internet]. Genebra: World Health trânsito na população brasileira, 2013. Ciên Saúde
Organization; 2018 [accessed on Aug 6, 2020]. Available Coletiva 2016; 21(12): 3777-86. http://dx.doi.
from: https://www.who.int/violence_injury_prevention/ org/10.1590/1413-812320152112.25692015
road_safety_status/2018/en/ 11. Brasil. Lei nº 11.705, de 19 de junho de 2008. Altera a
2. Wijnen W, Stipdonk H. Social costs of road crashes: Lei nº 9.503, de 23 de setembro de 1997, que ‘institui
An international analysis. Accid Anal Prev 2016; 94: o Código de Trânsito Brasileiro’, e a Lei nº 9.294, de
97-106. http://dx.doi.org/10.1016/j.aap.2016.05.005 15 de julho de 1996. Diário Oficial da União [Internet]
3. Instituto Brasileiro de Geografia e Estatística (IBGE). 20 jun. 2008 [accessed on Aug 6, 2020]. Available from:
Coordenação de Trabalho e Rendimento. Pesquisa http://www.planalto.gov.br/ccivil_03/_ato2007-
nacional de saúde: 2013: acesso e utilização dos serviços 2010/2008/lei/L11705.htm
de saúde, acidentes e violências: Brasil, grandes regiões 12. Malta DC, Soares Filho AM, Montenegro MMS,
e unidades da federação [Internet]. Rio de Janeiro: Mascarenhas MDM, Silva MMA, Lima CM, et al.
IBGE, 2015 [accessed on Aug 6, 2020]. Available from: Análise da mortalidade por acidentes de transporte
https://biblioteca.ibge.gov.br/visualizacao/livros/ terrestre antes e após a Lei Seca - Brasil, 2007-2009.
liv94074.pdf Epidemiol Serv Saúde 2010; 19(4): 317-28. http://
4. Campos MR, von Doellinger V dos R, Mendes LV, dx.doi.org/10.5123/S1679-49742010000400002
Costa M de F, Pimentel TG, Schramm JM. Morbidity 13. Andreuccetti G, Carvalho HB, Cherpitel CJ, Ye Y, Ponce
and mortality associated with injuries: results of the JC, Kahn T, et al. Reducing the legal blood alcohol
Global Burden of Disease study in Brazil, 2008. Cad concentration limit for driving in developing countries:
Saúde Pública 2015; 31(1): 121-36. http://dx.doi. a time for change? Results and implications derived
org/10.1590/0102-311x00191113 from a time series analysis (2001–2010) conducted
5. Calinescu T, Adminaite D. Progress in reducing drink in Brazil: Reducing blood alcohol limits for driving.
driving in Europe [Internet]. Bruxelas: European Addiction 2011; 106(12): 2124-31. http://dx.doi.
Safety Research Council, 2018 [accessed on Aug 6, org/10.1111/j.1360-0443.2011.03521.x
2020]. Available from: https://etsc.eu/wp-content/ 14. Abreu AMM, Jomar RT, Thomaz RGF, Guimarães
uploads/report_reducing_drink_driving_final.pdf RM, Lima JMB, Figueiró RFS. Impacto da lei seca na
6. Liu YC, Ho CH. Effects of different blood alcohol mortalidade por acidentes de trânsito. Rev Enferm
concentrations and post-alcohol impairment on UERJ 2012; 20(1): 21-6.
driving behavior and task performance. Traffic 15. Klabunde FC, Ghizzo Filho J, Freitas PF, Nazário
Inj Prev 2010; 11(4): 334-41. http://dx.doi. NO. Impacto da lei seca na taxa de mortalidade por
org/10.1080/15389581003747522 acidentes de trânsito, Santa Catarina, entre 2005 e
7. Zhao X, Zhang X, Rong J. Study of the Effects of 2011 [impact of alcohol prohibition in the death rate
Alcohol on Drivers and Driving Performance on from traffic accidents, in Santa Catarina, from 2005
Straight Road. Math Probl Eng 2014. http://dx.doi. to 2011]. Arq Catarin Med 2017; 46(2): 108-17.
org/10.1155/2014/607652 16. Abreu DROM, Souza EM, Mathias TAF. Impacto
8. Cherpitel CJ, Witbrodt J, Korcha RA, Ye Y, Monteiro do Código de Trânsito Brasileiro e da Lei Seca na
MG, Chou P. Dose-response relationship of alcohol mortalidade por acidentes de trânsito [Impact of the
and injury cause: effects of country-level drinking Brazilian Traffic Code and the Law Against Drinking
pattern and alcohol policy. Alcohol Clin Exp Res and Driving on mortality from motor vehicle accidents].
2019; 43(5): 850-6. http://dx.doi.org/10.1111/ Cad Saúde Pública 2018; 34(8): e00122117. http://
acer.13986 dx.doi.org/10.1590/0102-311x00122117
9. Taylor B, Rehm J. The relationship between 17. Brasil. TabNet Win32 3.0: Mortalidade [Internet].
alcohol consumption and fatal motor vehicle [accessed on Nov 10, 2017]. Available from: http://
injury: high risk at low alcohol levels. Alcohol tabnet.datasus.gov.br/cgi/deftohtm.exe?sim/cnv/
Clin Exp Res 2012; 36(10): 1827-34. http://dx.doi. obt10br.def
org/10.1111/j.1530-0277.2012.01785.x 18. Brasil. Projeção da população [Internet]. [accessed on
10. Damacena GN, Malta DC, Boccolini CS, Souza Nov 15, 2017]. Available from: https://www.ibge.gov.
Júnior PRB, Almeida WS, Ribeiro LS, et al. Consumo br/apps/populacao/projecao/
12
REV BRAS EPIDEMIOL 2021; 24: E210045Impact of the Dry Law on road traffic mortality in Brazilian states: an interrupted time series analysis
19. Brasil. Estatísticas: frota de veículos [Internet]. [accessed 28. Queiroz MS, Oliveira PC. Acidentes de trânsito:
on Nov 25, 2017]. Available from: https://antigo. uma visão qualitativa no Município de Campinas,
infraestrutura.gov.br/component/content/article/115- São Paulo, Brasil [Traffic accidents: a qualitative
portaldenatran/8552-estat%C3%ADsticas-frota-de- approach from Campinas, São Paulo, Brazil]. Cad Saúde
ve%C3%ADculos-denatran.html Pública 2002; 18(5): 1179-87. https://doi.org/10.1590/
20. Lefio A, Bachelet VC, Jiménez-Paneque R, Gomolán s0102-311x2002000500010
P, Rivas K. A systematic review of the effectiveness of 29. Teles AS, Coelho TCB, Ferreira MPS, Scatena
interventions to reduce motor vehicle crashes and their JHG. Serviço de Atendimento Móvel de Urgência
injuries among the general and working populations. (SAMU) do Estado da Bahia: subfinanciamento e
Rev Panam Salud Publica 2018; 42: e60. http://dx.doi. desigualdade regional [Mobile Emergency Care Service
org/10.26633/RPSP.2018.60 (SAMU): underfunding and regional inequality].
21. Carvalho CHR, Pereira RHM. Gastos das famílias brasileiras Cad Saúde Colet 2017; 25(1): 51-7. https://doi.
com transporte urbano público e privado no Brasil: uma org/10.1590/1414-462x201700010188
análise da POF 2003 e 2009 [Internet]. Brasília: Instituto 30. Machado CV, Salvador FG, O’Dwyer G. Mobile
de Pesquisa Econômica e Aplicada, 2012 [accessed on Emergency Care Service: analysis of Brazilian policy.
Aug 6, 2020]. Available from: https://www.ipea.gov.br/ Rev Saúde Pública 2011; 45(3): 519-28. https://doi.
portal/images/stories/PDFs/TDs/td_1803.pdf org/10.1590/s0034-89102011005000022
22. Pero V, Stefanelli V. A questão da mobilidade urbana nas 31. Sterne JAC, Hernán MA, Reeves BC, Savovic J,
metrópoles brasileiras. Rev Econ Contemp 2015; 19(3): Berkman ND, Viswanathan M, et al. ROBINS-I: a
366-402. http://dx.doi.org/10.1590/198055271932 tool for assessing risk of bias in non-randomised
23. Nascimento AS, Menandro PRM. Bebida Alcoólica studies of interventions. BMJ 2016; 355. https://
e Direção Automotiva: relatos de policiais militares doi.org/10.1136/bmj.i4919
sobre a “Lei Seca”. Psicol Ciênc Prof 2016; 36(2): 411- 32. Penfold RB, Zhang F. Use of interrupted time series
25. http://dx.doi.org/10.1590/1982-370300067201 analysis in evaluating health care quality improvements.
24. Almeida RL, Bezerra Filho JG, Braga JU, Magalhães Acad Pediatr 2013; 13(6 Supl.): S38-44. https://doi.
FB, Macedo MC, Silva KA. Man, road and vehicle: risk org/10.1016/j.acap.2013.08.002
factors associated with the severity of traffic accidents.
Rev Saude Publica 2013; 47(4): 718-31. https://doi. Received on: 01/22/2021
org/10.1590/S0034-8910.2013047003657 Revised on: 06/01/2021
25. Lima IMO, Figueiredo JC, Morita PA, Gold P. Fatores Approved on: 06/04/2021
condicionantes da gravidade dos acidentes de trânsito
nas rodovias brasileiras [Internet]. Brasília: Instituto Author contributions: H.R.C. Nunes participated
de Pesquisa Econômica e Aplicada; 2008 [accessed on in the conception and design of the study, statistical
Aug 6, 2020]. Available from: https://www.ipea.gov. analysis, data interpretation and writing of the
br/portal/images/stories/PDFs/TDs/td_1344.pdf manuscript. C. Murta-Nascimento contributed to
26. Brasil. Lei nº 9.503 de 23 de setembro de 1997. Institui data interpretation of and writing of the manuscript.
o Código Brasileiro de Trânsito. Brasil; 1997. M.C.P. Lima participated in the conception and design
27. Marques EV, Machado MA. Identificação dos fatores of the study, data interpretation and writing of the
relevantes na decisão da alocação dos recursos manuscript. All authors read and approved the final
econômicos visando um trânsito seguro. Rev Adm version of the paper. They declare that they are
Pública 2010; 44(6): 1379-404. https://doi.org/10.1590/ responsible for all aspects of the work, ensuring its
S0034-76122010000600006 accuracy and integrity.
© 2021 Associação Brasileira de Saúde Coletiva
This is an open access article distributed under the terms of the Creative Commons license.
13
REV BRAS EPIDEMIOL 2021; 24: E210045You can also read
